S-MDP: Streaming With Markov Decision Processes

Khan, Koffka; Goodridge, Wayne

doi:10.1109/tmm.2019.2892304

Cited by 22 publications

(18 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The pseudo-subjective evaluation method mainly uses artificial intelligence and statistical theory to build complex training models and service evaluation mechanisms for different types of video services through big data processing such as real-time video service quality evaluation based on RNN (Recurrent Neural Network) [19], OTT (Over The Top) video service quality evaluation based on idealized data clustering [20], network video quality evaluation based on fuzzy expert system [21], video service quality evaluation based on S-MDP (Streaming with Markov Decision Processes) [22], and mobile video quality evaluation based on wireless cellular network [23]. In general, although the pseudo-subjective evaluation method combines the advantages of subjective evaluation and objective evaluation to better ensure the real-time and accuracy of video service quality evaluation, the existing research results focus on the quality of service in the single-path transmission environment.…”

Section: Related Workmentioning

confidence: 99%

Research on the Load Distribution Strategy to Meet the QoE Requirements for Conversational Real-Time HD Video Service

et al. 2020

View full text Add to dashboard Cite

A reliable transmission with QoE (Quality of Experience) guarantee is crucial for internet conversational service applications. However, due to the limited network bandwidth bottleneck effect and the drawback of transmission technology, there exists no mature and open QoE technical solution for this service. In this paper, we focused our attention on a load distribution strategy for multipath relay transmission to meet the QoE requirements of conversational real-time HD video services. It consisted of three stages: First, a series of relay nodes was deployed in the backbone network, and a software defined overlay network was constructed to perform the multipath relay transmission for the service. Second, by an analysis of the QoE feature, a bijection was built for each quantitative QoE and its MOS (Mean Opinion Score) score. Finally, considering the influence of the coupling relation between paths on the service quality in multipath relay transmission, fuzzy cooperative game theory was used to design the service load distribution strategy. Many experiments showed that compared with state-of-the-art methods in the single-path transmission scene, the strategy we designed can dynamically adjust the load distribution of each sub-path according to the change in QoS (Quality of Service) information of the transmission path in real time. While meeting the strict real-time constraints of the service, it can effectively avoid the impact of network random congestion on the service QoE.

show abstract

Section: Related Workmentioning

confidence: 99%

Research on the Load Distribution Strategy to Meet the QoE Requirements for Conversational Real-Time HD Video Service

et al. 2020

View full text Add to dashboard Cite

show abstract

“…The concept of adaptive video streaming is based on the idea to adapt the bandwidth required by the video stream to the throughput available on the network path from the stream source to the client [32]. These algorithms can live at the server [26], at an intermediate network device [25] or at the client [27], [21]. With client-side protocols it is the player that decides what bitrate to request for any fragment, improving server-side scalability [1].…”

Section: Dash and Multiple Competing Players Problemmentioning

confidence: 99%

“…In the presence of competing HTTP-based adaptive streaming (HAS) clients the TCP throughput does not always faithfully represent the fair-share bandwidth [29]. Three performance issues that can take place when two or more adaptive streaming players share a network bottleneck and compete for available bandwidth are instability, unfairness and utilization [27]. It is shown that in the case of two competing video flows Adaptive video streaming players provide a received video rate that oscillates around the fair share, but with an increased number of video level switches [18].…”

Section: Dash and Multiple Competing Players Problemmentioning

confidence: 99%

“…Many DASH approaches require choosing the best option among a range of competing video segments [29], [27], [21], [18]. These so-called optimization problems can be solved using mathematical methods such as linear programming to evaluate maximum advantages or minimum expenses given some goals and under certain limitations for equilibrium-assumed deterministic systems.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Stochastic Dynamic Programming in DASH

Khan¹,

Goodridge²

2019

IJANA

Self Cite

View full text Add to dashboard Cite

Stochastic Dynamic Programming (SDP) is an important class of optimization which has recently been used in the area of dynamic adaptive streaming over HTTP (DASH). Though DASH is very popular method of video delivery in recent years it is plagued with problems when multiple players share a bottleneck link. Thus, this area has become a very active area of research. Two works which implement SDP in DASH are selected and their performances compared against the Conventional DASH player. It is shown that SDP works well for various network conditions.

show abstract

“…The shared bottleneck may be an edge network link in this scenario. It has been previously observed that such competition can lead to performance issues [2] [1] [16] [13]. The concept of adaptive video streaming (see Figure 1) is based on the idea to adapt the bandwidth required by the video stream to the throughput available on the network path from the stream source to the client [1].…”

Section: Introductionmentioning

confidence: 99%

Collaborative Methods to Reduce the Disastrous Effects of the Overlapping ON Problem in DASH

Khan¹,

Goodridge²

2019

IJANA

Self Cite

View full text Add to dashboard Cite

The performance of today's adaptive video streaming players (DASH) is severely hindered by overlapping ON-OFF patterns that occurs during a streaming session. High switching rates, freezing and skipping annoy users resulting in poor user-QoE. This paper attempts to overcome the ON-OFF problem by keeping players aware of each other's downloads and inter-request times. Using this a player is able to better predict future player actions and reschedule their downloads to produce least conflicts with others. If a player's start segment download overlaps with the end of one or more players' current downloads within a certain time t 1 it waits until the download of others completes before starting its own download. Conversely if too many players are sharing the bottleneck link a player will hold of its download by a time t 2 enabling others to move towards completion of their downloads. This reduces the overlap of its download with other players. Both these methods help reduce competition at the bottleneck and produces a more balanced sharing of network resources.

show abstract

S-MDP: Streaming With Markov Decision Processes

Cited by 22 publications

References 26 publications

Research on the Load Distribution Strategy to Meet the QoE Requirements for Conversational Real-Time HD Video Service

Research on the Load Distribution Strategy to Meet the QoE Requirements for Conversational Real-Time HD Video Service

Stochastic Dynamic Programming in DASH

Collaborative Methods to Reduce the Disastrous Effects of the Overlapping ON Problem in DASH

Contact Info

Product

Resources

About